Gert Jan Boender

Risk Maps for the Spread of Highly Pathogenic Avian Influenza in Poultry

Devastating epidemics of highly contagious animal diseases such as avian influenza, classical swine fever, and foot-and-mouth disease underline the need for improved understanding of the factors promoting the spread of these pathogens. Here the authors present a spatial analysis of the between-farm transmission of a highly pathogenic H7N7 avian influenza virus that caused a large epidemic in The Netherlands in 2003. The authors developed a method to estimate key parameters determining the spread of highly transmissible animal diseases between farms based on outbreak data. The method allows for the identification of high-risk areas for propagating spread in an epidemiologically underpinned manner. A central concept is the transmission kernel, which determines the probability of pathogen transmission from infected to uninfected farms as a function of interfarm distance. The authors show how an estimate of the transmission kernel naturally provides estimates of the critical farm density and local reproduction numbers, which allows one to evaluate the effectiveness of control strategies. For avian influenza, the analyses show that there are two poultry-dense areas in The Netherlands where epidemic spread is possible, and in which local control measures are unlikely to be able to halt an unfolding epidemic. In these regions an epidemic can only be brought to an end by the depletion of susceptible farms by infection or massive culling. The analyses provide an estimate of the spatial range over which highly pathogenic avian influenza viruses spread between farms, and emphasize that control measures aimed at controlling such outbreaks need to take into account the local density of farms.

Quantitative analysis of transmission parameters for bluetongue virus serotype 8 in Western Europe in 2006

The recent bluetongue virus serotype 8 (BTV-8) epidemic in Western Europe struck hard. Controlling the infection was difficult and a good and safe vaccine was not available until the spring of 2008. Little was known regarding BTV transmission in Western Europe or the efficacy of control measures. Quantitative details on transmission are essential to assess the potential and efficacy of such measures.To quantify virus transmission between herds, a temporal and a spatio-temporal analysis were applied to data on reported infected herds in 2006. We calculated the basic reproduction number between herds (Rh: expected number of new infections, generated by one initial infected herd in a susceptible environment). It was found to be of the same order of magnitude as that of an infection with Foot and Mouth Disease (FMD) in The Netherlands, e.g. around 4. We concluded that an average day temperature of at least 15°C is required for BTV-8 transmission between herds in Western Europe. A few degrees increase in temperature is found to lead to a major increase in BTV-8 transmission.We also found that the applied disease control (spatial zones based on 20 km radius restricting animal transport to outside regions) led to a spatial transmission pattern of BTV-8, with 85% of transmission restricted to a 20 km range. This 20 km equals the scale of the protection zones. We concluded that free animal movement led to substantial faster spread of the BTV-8 epidemic over space as compared to a situation with animal movement restrictions.

Transmission risks and control of foot-and-mouth disease in The Netherlands: spatial patterns.

In 2001 the epidemics of foot-and-mouth disease virus (FMDV) in Great Britain, The Netherlands and France have shown how fast FMDV may spread between farms. The massive socio-economic impact of these epidemics and the intervention measures taken demonstrate the need for quantitative assessments of the efficacy of candidate intervention strategies. Here we use a mathematical model to describe the spatial transmission of FMDV in The Netherlands and use the Dutch 2001 outbreak data to estimate model parameters. We assess the effect of ring culling strategies using a novel and fast approach producing risk maps, and discuss its consequences for ring vaccination. These risk maps identify both the geographical areas of low risk, where a given intervention strategy is likely to achieve epidemic control within only two or three farm-to-farm infection generations, and high-risk areas, where control is likely to take (much) longer. Our results indicate that certain densely populated livestock areas in the Netherlands remain high-risk areas even for strategies that extend EU minimum measures with culling or vaccination within a ring radius of several kilometres. Depending on an economic assessment, area-wide vaccination might be judged appropriate once an FMDV outbreak would have been confirmed in or close to such a high-density area. The modeling approach developed here could be readily applied to outbreak data for other diseases and in other countries.

Predicting future costs of High-Pathogenicity Avian Influenza epidemics: large versus small uncertainties

Every five years, the Dutch government and the poultry sector agree on how the direct costs of epidemics in poultry, should they occur, will be shared. In the agreement for 2005–2009 the maximum amount to be paid by the poultry sector was set considerably higher than in the 1999–2004 agreement. This increase was caused mainly by the expected financial risks associated with High-Pathogenicity Avian Influenza (HPAI) epidemics. In this paper we focus on elucidating the uncertain and the less uncertain aspects of the HPAI financial risk problem. We distinguish between (1) the probability of an introduction of HPAI in the Netherlands, (2) the transmission potential of HPAI in the Netherlands, and (3) the costs and financing issues resulting from HPAI epidemics. We argue that whereas current understanding allows relatively precise answers to the question ‘If there is an there is an epidemic, how many farms will be affected and what will be the direct costs?’, much larger uncertainties are associated with the questions ‘What is the chance of an HPAI epidemic in the Netherlands?’, ‘How large will be the long-term government share in the direct costs?’, and ‘How large will be the indirect costs?’.

The Influence of Between-Farm Distance and Farm Size on the Spread of Classical Swine Fever during the 1997–1998 Epidemic in The Netherlands

As the size of livestock farms in The Netherlands is on the increase for economic reasons, an important question is how disease introduction risks and risks of onward transmission scale with farm size (i.e. with the number of animals on the farm). Here we use the epidemic data of the 1997–1998 epidemic of Classical Swine Fever (CSF) Virus in The Netherlands to address this question for CSF risks. This dataset is one of the most powerful ones statistically as in this epidemic a total of 428 pig farms where infected, with the majority of farm sizes ranging between 27 and 1750 pigs, including piglets. We have extended the earlier models for the transmission risk as a function of between-farm distance, by adding two factors. These factors describe the effect of farm size on the susceptibility of a ‘receiving’ farm and on the infectivity of a ‘sending’ farm (or ‘source’ farm), respectively. Using the best-fitting model, we show that the size of a farm has a significant influence on both farm-level susceptibility and infectivity for CSF. Although larger farms are both more susceptible to CSF and, when infected, more infectious to other farms than smaller farms, the increase is less than linear. The higher the farm size, the smaller the effect of increments of farm size on the susceptibility and infectivity of a farm. Because of changes in the Dutch pig farming characteristics, a straightforward extrapolation of the observed farm size dependencies from 1997/1998 to present times would not be justified. However, based on our results one may expect that also for the current pig farming characteristics in The Netherlands, farm susceptibility and infectivity depend non-linearly on farm size, with some saturation effect for relatively large farm sizes.

Confirmation of spatial patterns and temperature effects in Bluetongue virus serotype-8 transmission in NW-Europe from the 2007 reported case data

Two separate analyses were carried out to understand the epidemiology of Bluetongue virus serotype 8 (BTV-8) in 2007 in North West Europe: First, the temporal change in transmission rates was compared to the evolution of temperature during that season. Second, we evaluated the spatio-temporal dynamics of newly reported outbreaks, to estimate a spatial transmission kernel. For both analyses, the approach as used before in analysing the 2006 BTV-8 epidemic had to be adapted in order to take into account the fact that the 2007 epidemic was not a newly arising epidemic, but one advancing from whereto it had already spread in 2006. We found that within the area already affected by the 2006 outbreak, the pattern of newly infected farms in 2007 cannot be explained by between-farm transmission, but rather by local re-emergence of the virus throughout that region. This indicates that persistence through winter was ubiquitous for BTV-8. Just like in 2006, we also found that the temperature at which the infection starts to spread lies close to 15 °C. Finally, we found that the shape of the transmission kernel is in line with the one from the 2006 epidemic. In conclusion, despite the substantial differences between 2006 and 2007 in temperature patterns (2006 featured a heat wave in July, whereas 2007 was more regular) and spatial epidemic extent, both the minimum temperature required for transmission and the transmission kernel were similar to those estimated for the 2006 outbreak, indicating that they are robust properties, suitable for extrapolation to other years and similar regions.

Spatial transmission of H5N2 highly pathogenic avian influenza between Minnesota poultry premises during the 2015 outbreak

The spatial spread of highly pathogenic avian influenza (HPAI) H5N2 during the 2015 outbreak in the U.S. state of Minnesota was analyzed through the estimation of a spatial transmission kernel, which quantifies the infection hazard an infectious premises poses to an uninfected premises some given distance away. Parameters were estimated using a maximum likelihood method for the entire outbreak as well as for two phases defined by the daily number of newly detected HPAI-positive premises. The results indicate both a strong dependence of the likelihood of transmission on distance and a significant distance-independent component of outbreak spread for the overall outbreak. The results further suggest that HPAI spread differed during the later phase of the outbreak. The estimated spatial transmission kernel was used to compare the Minnesota outbreak with previous HPAI outbreaks in the Netherlands and Italy to contextualize the Minnesota transmission kernel results and make additional inferences about HPAI transmission during the Minnesota outbreak. Lastly, the spatial transmission kernel was used to identify high risk areas for HPAI spread in Minnesota. Risk maps were also used to evaluate the potential impact of an early marketing strategy implemented by poultry producers in a county in Minnesota during the outbreak, with results providing evidence that the strategy was successful in reducing the potential for HPAI spread.

Associations between pneumonia and residential distance to livestock farms over a five-year period in a large population-based study

In a recent study of electronic health records (EHR) of general practitioners in a livestock-dense area in The Netherlands in 2009, associations were found between residential distance to poultry farms and the occurrence of community-acquired pneumonia (CAP). In addition, in a recent cross-sectional study in 2494 adults in 2014/2015 an association between CAP and proximity to goat farms was observed. Here, we extended the 2009 EHR analyses across a wider period of time (2009–2013), a wider set of health effects, and a wider set of farm types as potential risk sources. A spatial (transmission) kernel model was used to investigate associations between proximity to farms and CAP diagnosis for the period from 2009 to 2013, obtained from EHR of in total 140,059 GP patients. Also, associations between proximity to farms and upper respiratory infections, inflammatory bowel disease, and (as a control disease) lower back pain were analysed. Farm types included as potential risk sources in these analyses were cattle, (dairy) goats, mink, poultry, sheep, and swine. The previously found association between CAP occurrence and proximity to poultry farms was confirmed across the full 5-year study period. In addition, we found an association between increased risk for pneumonia and proximity to (dairy) goat farms, again consistently across all years from 2009 to 2013. No consistent associations were found for any of the other farm types (cattle, mink, sheep and swine), nor for the other health effects considered. On average, the proximity to poultry farms corresponds to approximately 119 extra patients with CAP each year per 100,000 people in the research area, which accounts for approximately 7.2% extra cases. The population attributable risk percentage of CAP cases in the research area attributable to proximity to goat farms is approximately 5.4% over the years 2009–2013. The most probable explanation for the association of CAP with proximity to poultry farms is thought to be that particulate matter and its components are making people more susceptible to respiratory infections. The causes of the association with proximity to goat farms is still unclear. Although the 2007–2010 Q-fever epidemic in the area probably contributed Q-fever related pneumonia cases to the observed additional cases in 2009 and 2010, it cannot explain the association found in later years 2011–2013.

Spatial transmission of Swine Vesicular Disease virus in the 2006-2007 epidemic in Lombardy.

In 2006 and 2007 pig farming in the region of Lombardy, in the north of Italy, was struck by an epidemic of Swine Vesicular Disease virus (SVDV). In fact this epidemic could be viewed as consisting of two sub-epidemics, as the reported outbreaks occurred in two separate time periods. These periods differed in terms of the provinces or municipalities that were affected and also in terms of the timing of implementation of movement restrictions. Here we use a simple mathematical model to analyse the epidemic data, quantifying between-farm transmission probability as a function of between-farm distance. The results show that the distance dependence of between-farm transmission differs between the two periods. In the first period transmission over relatively long distances occurred with higher probability than in the second period, reflecting the effect of movement restrictions in the second period. In the second period however, more intensive transmission occurred over relatively short distances. Our model analysis explains this in terms of the relatively high density of pig farms in the area most affected in this period, which exceeds a critical farm density for between-farm transmission. This latter result supports the rationale for the additional control measure taken in 2007 of pre-emptively culling farms in that area.